Utilizing Range-Doppler Characteristics for Classifying Insect and Bird Echoes in Weather Radar

The global climate change has led to a sharp decline in the number and species diversity of aerial migratory animals. Breakthroughs in the field of ecological monitoring using weather radar enable large-scale and long-term ecological monitoring. However, the primary challenges are the compression of spectral details in base data products, which leads to a loss of scatterer-information, and bias caused by frequency offset in dual-polarized products, which affects classification consistency between radar sites. To address these challenges, we explored the multi-dimensional range-Doppler (RD) characteristics of migration traits from Level-I IQ data and proposed an echo classification method for insect and bird of weather radar. We employ adaptive linear filtering for clutter preprocessing, followed by morphological image process to extract biological connected domains, leveraging spectral feature differences between insects and birds. Subsequently, a hierarchical classifier model is developed for classification, complemented by a minimal value inflection points detection method to identify insect-bird coexistence. Our approach is implemented to support the large-scale monitoring of aerial animal migration in the network of S-band weather radar stations. Experiments conducted with five operational weather radars have comprehensively validated the benefits of the proposed method in the accurate classification of insect and bird echoes. Future work will concentrate on precise species identification and biological quantification within resolution volumes.